Construction and Building Materials 25 (2011) 41214125

Contents lists available at ScienceDirect

Construction and Building Materials

journal homepage: www.elsevier.com/locate/conbuildmat

Gypsum hemihydratecement blends to improve renderings durability

Gladis Camarini a,, Jos Antonio De Milito b,1
a
GEDRRIC, LMC, LARES, University of Campinas, Department of Architecture and Construction, School of Civil Engineering, Architecture and Urban Design,
Cidade Universitria Zeferino Vaz, s/n, CP 6021, CEP 13083-970, Campinas, Brazil
b
Faculty of Engineering of Sorocaba FACENS, Rodovia Senador Jos Ermrio de Moraes, 1425, Castelinho km 1,5 - Alto da Boa Vista, CEP 18087-125, Sorocaba, Brazil

a r t i c l e i n f o a b s t r a c t

Article history: Gypsum is a historical binder that was already used thousands of years ago. More recently its employ-
Received 21 September 2010 ment in construction had a signicant growth. It is a binding material being used in masonry as a plaster
Received in revised form 8 April 2011 nish coat. Despite being economical, having good performance and easy application in buildings, its use
Accepted 14 April 2011
is limited to indoors due to its solubility. Some efforts have been made to improve its performance in
Available online 6 May 2011
external environment. In that sense, researches advance in the attempt at using several admixtures like
epoxy resins, polymers, cements, slags and waterproong materials. The aim of this work was to inves-
Keywords:
tigate the effect of blastfurnace slag Portland cement addition to b-hemihydrate and observe the changes
Gypsum
Blastfurnace slag Portland cement
in mechanical properties and microstructure to improve gypsum performance. Results show an improve-
b-Hemihydrate ment of gypsum performance when this kind of cement is added to the mixture and a 3-years rendering
Ettringite formation exposed to external environment had no changes, indicating that there was no ettringite formation. SEM
Renderings performance analysis showed a more compact structure which can result in a water resistant composite.
2011 Elsevier Ltd. All rights reserved.

1. Introduction The reactivity of gypsum-based binders is produced by gypsum

Gypsum is a calcium sulfate dihydrate (CaSO42H2O) which oc-
curs in different countries and its use is widespread all over the
world. It can be used in both natural and burnt forms. Natural gyp-
sum is used in agriculture and in Portland cement plants. Burnt
gypsum is used to produce hemihydrate, extensively used in build-
ings, ceramics and medical industries.
In building construction, gypsum became a popular nishing
material due to its excellent performance, attractive appearance,
easy application, and its healthful contribution to living conditions
[1]. This building material is abundant in Brazil and has high pur-
ity. The natural gypsum rock reserves are estimated to be about
1.6 billion tons and the gypsum production in 2007 was almost
2 millions tons, and 89% of this production comes from Pernam-
buco State at Northest region [2]. Availability, the relatively low
level of start-up investments, and a favorable market situation,
all provide conditions for growth and the protable industrial pro-
duction of gypsum-based materials. The majority of gypsum pro-
duction, 59% is burnt, 30% consumed by cement industry, and
(dihydrate) (CaSO42H2O) dehydration. Depending on dehydration
temperature, vapor pressure inside of furnace, heating control,
neness and density of the raw-material it is obtained different
products. Gypsum looses water when burnt at different tempera-
tures as follows: b-hemihydrate (CaSO40.5H2O) is formed when
gypsum is heated at temperatures between 100 C and 180 C.
Anhydrite III (CaSO4eH2O) (0.06 < e < 0.11) is formed when gyp-
sum is heated at temperatures below 300 C. Anhydrite II (CaSO4)
is formed when gypsum is heated at temperatures higher than
300 C. Inside the furnace the temperature is not totally uniform
and the commercial plaster for building construction can have both
products: hemihydrate (HH) and anhydrite III (AH-III) [3]. Reac-
tions (1)(3) show these dehydrations.
<180  C
CaSO4  2H2 O ! CaSO4  0:5H2 O 1:5H2 O 1
Gypsum b-hemihydrate
<300  C
CaSO4  2H2 O ! CaSO4  eH2 O 2  eH2 O 2
Gypsum Anhydrite III

11% by the agriculture. It is estimated that hemihydrate consump-

tion is by plates industry, rendering, molds, ceramics and others >300  C
CaSO4  2H2 O ! CaSO4 2H2 O 3
[2]. Gypsum Anhydrite II

Both b-hemihydrate and anhydrite III at normal temperature re-

Corresponding author. Tel.: +55 19 35 21 23 61; fax: +55 19 35 21 24 11. act with water forming dihydrate (gypsum). This reaction makes a
E-mail addresses: gcamarini@gmail.com, camarini@fec.unicamp.br (G. Camarini),
network of gypsum crystals that produces the properties of gyp-
milito@facens.br (J.A. De Milito). sum-based materials. This reaction releases heat that must be con-
1
Tel.: +55 15 32 38 11 88. trolled to avoid fast setting [4].

0950-0618/$ - see front matter 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.conbuildmat.2011.04.048
4122 G. Camarini, J.A. De Milito / Construction and Building Materials 25 (2011) 41214125

In cement, gypsum is added to clinker Portland to control set- Table 2

ting times. This addition is important to cement because it keeps Some works on gypsum-based binders for improved water resistance.

the cement in plastic state during the necessary period of time to Year Author Gypsum-based-binder
maintain cement composites workability. Without gypsum ground 1969 Sherr and Roshal [8] Gypsumacrylic polymers
clinker exhibits ash set, and in a few minutes makes the cement 1992 Singh and Garg [11] Phosphogypsumy ashblastfurnace
matrix stiff with no chances for remixing. For this reason there is slagOPC
no commercial Portland cement without gypsum addition [5]. 1994 Bentur et al. [12] PlasterOPCsilica fume
1995 Singh and Garg [9] Phosphogypsumy ashhydrated
When Portland cement is mixed with water a compound called limeOPC
ettringite is immediately formed. It is formed within hours and 1996 Singh and Garg [13] Phosphogypsumblastfurnace slag
does not cause any damaged action. The setting control has been OPC
attributed to this early ettringite formation (EEF). This compound 1996 Alsadi et al. [14] Gypsumgum Arabiccalcium
hydroxide
is also called primary ettringite [6] (Reaction (4)).
1996 Roy et al. [15] PhosphogypsumOPC
1997 Dharakumar et al. [16] Plasterepoxy resin
1998 Yan and You [17] Fluorgypsumy ashOPC
3CaO  Al2 O3 3CaSO4  2H2 O 26H2 O 1998 Klover [18,19] GypsumOPCsilica fume
1999 Yan et al. [20] Phosphogypsumy ashOPC
! Ca6 AlOH6 2  SO4 2  26H2 O 4
1999 Sayil and olak [21] Gypsumpolymer
When ettringite forms later, after several months or years, it is 2000 Singh [22] Phosphogypsumuorgypsum
natural gypsumcement clinker
called delayed etringite formation (DEF). The related heteroge- blastfurnace slag
neous expansion in a very rigid hardened material which can pro- 2001 olak [23] Gypsumacrylic latexepoxy resin
duce cracking and spalling. The disruptive effect is due to the 2001 Klover [24] GypsumOPCsilica fume
nonuniform expansion localized only in the area of the material 2001a De Milito [3] Gypsumblastfurnace slag Portland
cement
where ettringite forms. The source of sulfate can be internal (gyp-
2002 olak [25] GypsumOPCnatural pozzolan
sum) or external (environment) (Table 1) [6]. This is the reason 2002 Arikan and Sobolev [1] Gypsumpolymersadmixtures
why b-hemihydrate and cement blends are not so common and 2006 olak [26] Plasterpolymer
it is been subject of some researches. 2007 Shen et al. [27] Phosphogypsumy ashlime
In order to improve water resistance of gypsum-based materi- 2007a De Milito [28] Gypsumblastfurnace slag Portland
cementsilica fume
als, some efforts were made in this issue. In that sense, research ad- 2008 Demir and Baspinar [29] Gypsumy ashlime
vance in attempt at using: several admixtures like epoxy resins [7],
a
polymers [8], slags [9], and waterproong materials [10]. Brazilian works about this subject.

Table 2 shows some of these experiences. The works which

used Portland cement employed an ordinary Portland cement
(OPC).
In Brazil, the rst attempt to use gypsumcement blends was in
1980 in precast panels for walls. These walls would be used in
low-income housing projects. During the construction period
(19801983) the Construction Company had a contract to build
3.000 residential units. This Company used a non-tested product
called chemical plastic concrete [30]: a mixture of OPC, b-hemi-
hydrate, aggregates, admixtures and water. The binder composi-
tion was 60% of b-hemihydrate and 40% of OPC [28]. These
panels were used to internal and external walls. Before ending
the construction of all units, the rst ones were already deterio-
rated due to delayed ettringite formation (Fig. 1).
After this experience, our research group began to study some
mixture designs between b-hemihydrate and blastfurnace slag
Portland cement and the results were very promising. Some of
the results are presented in this work.
The present work investigated the performance of b-hemihy-
drate mixed with blastfurnace slag Portland cement to be used
for renderings when applied to internal and external environ-
ments. In order to verify the performance of these mixtures,
3 years old wall renderings were observed. With the results it
was possible to answer the question about the possibility of using
the mixture of these two binders which, in fact, is not usual due to
delayed ettringite formation.
2. Experimental program
2.1. Materials
The materials used in this experimental work were: blastfurnace slag Portland
cement (named CPIII RS) and commercial b-Hemihydrate. The properties of binders
used in this work are detailed in Tables 3 and 4. All the tests were performed in
accordance with Brazilian Standards.

2.2. Mix design and curing

Table 1
Delayed ettringite formation by external and internal sulphate attack [6].
Delayed ettringite formation
External sulphate attack
It occurs when environmental
sulphate (from water or soil)
penetrates the material in service
It occurs in a permeable material
It occurs in a moist environment
favoring diffusion of SO
4 through
the aqueous phase of the capillary
pores
b-Hemihydrate (b-HH) and b-hemihydrate-cement (b-HH-BFSC) mixtures stud-
ied in this work were: (i) 100% b-HH named A1-ext, cured in external environment,
and A4-int cured in internal environment, and (ii) blends of 75% of b-HH plus 25% of
Internal sulphate blastfurnace slag Portland cement (BFSC), by weight, named A2-ext those blends
cured in external environment and A5-int those blends cured in internal environ-
It occurs in a sulphate-free
ment [28].
environment for the late sulphate
Pastes A1-ext and A4-int had the same mixing proportions, which were 1: 0.70
release from gypsum-contaminated
(1 part of b-HH, and water/binder ratio 0.70).
aggregates or thermal decomposition
Pastes A2-ext and A5-int had the same mixing proportions and they were
of ettringite
1:0.34:0.70 (1 part of b-HH, 0.34 part of BFSC, and water/binder ratio 0.70), by
It is favored by preliminary
weight. The mixture was hand mixing for 2 min and after that the paste was used
microcracks where deposition of
to specimens production. The designations of the pastes, the water/binder ratio
ettringite crystals can occur
(w/b), and curing conditions of pastes are listed in Table 5.
It occurs in a moist environment
favoring diffusion of SO4 and other
reacting ions (Ca++ and aluminate) 2.3. Test methods
through water-saturated capillary
pores The pastes were prepared for compressive strength tests, SEM images and wall
renderings. The specimens for compressive strength test were kept in cylindrical
 G. Camarini, J.A. De Milito / Construction and Building Materials 25 (2011) 41214125 4123

Fig. 1. External walls deterioration due to ettringite formation [28].

Table 3 3. Results and discussion

Chemical composition, physical and mechanical properties of b-hemihydrate.

Chemical composition (%) Physical properties 3.1. Compressive strength

3
Free water 1.38 Bulk density 646 kg/m
Combined water 5.07 Fineness modulus 12.65
Fig. 2 shows the compressive strength results of the pastes in
IR (IR + SiO2) 0.64 external and internal environment curing condition. It can be seen
R2O3 0.19 Setting time (min) on results of b-HH pastes, exposed in internal environment (A4-
CaO 38.2 Initial 8 int), that compressive strength reach its maximum value at
MgO 0.48 Final 18
28 days and it does not increase with time. On the other hand,
SO3 54.5 Water/binder ratio 0.52
CO2 1.28 the strengths of b-HH pastes cured in external environment (A1-
CaSO40.5H2O 81.8 ext) diminished signicantly with time due to the action of water
CaSO4 13.8 Mechanical properties on the specimens, leaving the internal structure more porous due
CaCO3 1.72 Compressive strength (MPa)
to solubility of gypsum and its low water resistance.
MgCO3 1.00 7 days 17.5 MPa
IR + SiO2 + R2O3 0.83 Hardness (7 days) 22.8 N/mm2
The pastes with b-HH-BFSC blends showed differences in per-
SO3 (excess) 1.25 formance related to pure b-HH pastes. Surprisingly, compressive
strength of A5-int pastes decreased with time; results that were
not expected because they were in a protected environment with
no water access. This result must be more investigated. But micro-
Table 4
structure observations can explain it. In spite of that, in external
Chemical composition, physical, and mechanical properties of blastfurnace slag
Portland cement. environment A2-ext pastes showed a signicant difference in com-
pressive strength relate to the performance of pure gypsum (A1-
Chemical composition (%) Mineralogical composition (%)
ext). The performance of A2-ext paste is similar to the performance
Additions 62 C3S 21.0 of A4-int (pure b-HH paste) in internal environment. Thus, these
LOI 3.01 C2S 5.0
blends of b-HH-BFSC had good water resistance and showed good
RI 0.75 C3A 3.2
SiO2 25.93 C4AF 3.5 compressive strength results on this period.
Al2O3 8.38 The 3 years exposed wall renderings had a good performance
Fe2O3 2.09 Physical properties and there were no cracking and no spalling on their surface. This
CaO 51.53 Density (kg/m3) 3032 is a good result, indicative of no delayed ettringite formation.
MgO 3.78 Blaine neness (m2/kg) 3879
SO3 3.68
Na2O 0.14 Compressive strength (MPa)
K2O 0.55 3 days 20.5 3.2. Crystal morphology of the pastes
7 days 30.1
28 days 42.9 The microstructure of both mixtures exposed to internal and
external environment curing conditions is not similar. Fig. 3 shows

molds at 20 C 2 C for 24 h. After demolding, they were stored in internal (in-

doors) and external (outdoors) environment in order to simulate a real condition
Compressive strength
6
Compressive strength (MPa)

of exposition.
The compressive strength was determined at the ages of 28 days (1 month),
5
180 days (6 months) and 1080 days (3 years). Panels made with the same pastes
were exposed for a period of 3 years.
4
Microscopic examinations were carried out on fresh fractures of specimens
from compressive strength test using a Scanning Electron Microscope (SEM)
3
(JSM-5900LV). The SEM observations were made at Brazilian Syncrotron Light Lab-
oratory (LNLS), Laboratory of electronic Microscopy (LME).
2

Table 5 1
Mix proportions and curing conditions.
0
Mixture b-HH (%) BFSC (% by weight) w/b Curing condition A1 - ext A2 - ext A4 - int A5 - int
A1-ext 100 0 0.70 External environment Experimental mixtures
A4-int 100 0 0.70 Internal environment
28d 180s 1080d
A2-ext 75 25 0.70 External environment
A5-int 75 25 0.70 Internal environment
Fig. 2. Compressive strength results.
4124 G. Camarini, J.A. De Milito / Construction and Building Materials 25 (2011) 41214125
Fig. 3. SEM images of gypsum pastes exposed to (a) external (A1-ext) and (b) internal (A4-int) environments.
Fig. 4. SEM images of b-hemihydrate-cement blends pastes exposed to (a) external (A2-ext) and (b) internal (A5-int) environments.
SEM images of gypsum pastes exposed to external environment
(Fig. 3a) and internal environment (Fig. 3b).
In internal environment it can be seen a high degree of dihy-
drate crystal needles interlocked with few voids, and crystals of
low dimensions. This can explain the good compressive strength
performance. Pastes on external environment show an open struc-
ture with voids and longer crystal needles. This led to a worse per-
formance with lower compressive strength results.
Fig. 4 shows SEM images of b-HH-BFSC blends. Fig. 4a shows the
gypsumcement blend images in external environment (A2-ext)
and Fig. 4b shows blend image in internal environment (A5-int).
The structure of the paste A2-ext shows that the needles of
dihydrate crystals are engulfed by cement hydration products
(CSH) interlocking a cementdihydrate matrix which prevent
the dissolution of dihydrate and makes the structure stronger
and water resistant [24]. In internal environment (Fig. 4b) the nee-
dle crystals of dihydrate are not totally covered by cement hydra-
tion products (CSH). Some dihydrate needles are evident and
they are less resistant than cement hydration products, leaving
the binder structure more fragile. This can explain the compressive
strength performance diminishing with time.
4. Conclusions
The ndings of this investigation are summarized as follows:
1. The strength of the HHPC blends in external environment after
3 years of exposition is similar to HH in internal environment.
This result indicates the improved performance of such a blend
at longer ages. They also offer much better durability perfor-
mance than the gypsum dihydrate.
2. The b-HH-BFSC blends, using blastfurnace slag Portland cement,
improved the resistance of the pastes to water. Compressive
strength result is not as high as expected, but there is much bet-
ter performance than pure gypsum.
3. The microstructure of b-HH-BFSC blends are more compact and
a CSH lm covers the gypsum dihydrate needles, protecting
them against water attack.
4. After 3 years of exposition the wall renderings are still in a good
appearance, with neither cracks nor voids due to dihydrate dis-
solution, and showed no ettringite formation.
5. From these initial ndings, the question gypsum hemihy-
dratecement blends: are they possible? can be answered.
For the conditions employed in this work and using blastfur-
nace slag cement it is possible to do it. But many questions
must still be answered. For this, additional studies must be
performed in order to conrm these results and to solve both
theoretical (understanding setting, hydration mechanisms and
use of pozzolans on these blends) and applied questions (nd-
ing ways to improve water resistance of gypsum-based
binders).
Acknowledgments
The authors wish to thank the staff of Materials Construction
(LMC) and Binders and Waste Laboratories (LARES) for providing
the assistance in the mechanical tests. The authors are grateful
for given access to the facilities to Electronic Microscopy Labora-
tory (LME) from Brazilian Synchrotron Light Laboratory (LNLS)
for the SEM micrographs (proposals SEM-LV 2524 and SEM-LV
6247). The authors also thank CNPq research foundation (Process
No. 307049/2009-4).
 G. Camarini, J.A. De Milito / Construction and Building Materials 25 (2011) 41214125 4125

References [17] Yan P, You Y. Studies on the binder of y ashuorgypsumcement. Cem Concr
Res 1998;28(1):13540. doi:10.1016/S0008-8846(97)00247-0.
[18] Klover K. Setting and hardening of gypsumPortland cementsilica fume
[1] Arikan M, Sobolev K. The optimization of a gypsum-based composite material.
blends, part 1: temperature and setting expansion. Cem Concr Res
Cem Concr Res 2002;11(32):17258. doi:10.1016/S0008-8846(02)00858-X.
1998;28(3):42337. doi:10.1016/S0008-8846(98)00005-2.
[2] DNPM. National Department of Mineral Production, Gipsita; 2007. <http://
[19] Klover K. Setting and hardening of gypsumPortland cementsilica fume
www.dnpm-pe.gov.br/Detalhes/Gipsita.htm> [assessed 15.05.09].
blends, part 2: early strength, DTA, DRX and SEM observations. Cem Concr Res
[3] De Milito JA. Evaluation of gypsumPortland cementsilica fume pastes
1998;4(28):52331. doi:10.1016/S0008-8846(98)00004-0.
performance for rendering. Master of Science thesis, University of-
[20] Yan P et al. Microstructure and properties of the binder of y ash
Campinas, Brazil; 2001. 228p. <http://cutter.unicamp.br/document/?code=-
uorgypsumPortand cement. Cem Concr Res 1999;3(29):34954. doi:
vtls000227560> [in Portuguese].
10.1016/S0008-8846(98)00214-2.
[4] Maksoud MA, Ashour A. Heat of hydration as a method for determining the
[21] Sayil B, olak A. Durability of polymer modied and impregnated gypsum. In:
composition of calcined gypsum. Thermochim Acta 1981;3(46):3038.
Lacasse MA, Vanier DJ, editors. Durability of building materials and
doi:10.1016/0040-6031(81)80329-2.
components: Proceedings of the eighth international conference on
[5] Bhanumathidas N, Kalidas N. Dual role of gypsum: set retarder and strength
durability of building materials and components, vol. 8. Ottawa (Canada):
accelerator. Indian Concr J 2004;3(78):1703.
NRC Research Press; 2001.
[6] Collepardi M. A state-of-the-art review on delayed ettringite attack on
[22] Singh M. Inuence of blended gypsum on the properties of Portland cement
concrete. Cem Concr Compos 2003;45(25):4017. doi:10.1016/S0958-
and Portland slag cement. Cem Concr Res 2000;8(30):11858. doi:10.1016/
9465(02)00080-X.
S0008-8846(00)00306-9.
[7] olak A. Physical, mechanical, and durability properties of gypsumPortland
[23] olak A. Characteristics of acrylic latex-modied and partially epoxy-
cementnatural pozzolan blends. Can J Civil Eng 2001;28(3):37582.
impregnated gypsum. Cem Concr Res 2001;11(31):153947. doi:10.1016/
[8] Sherr AE, Roshal J. Plaster compositions: reduction in water absorption of
S0008-8846(01)00575-0.
gypsum plaster with acrylic polymers. I&EC Prod Res Dev 1969;8(2):1936.
[24] Klover K. Enhancing water resistance of cement and gypsumcement
[9] Singh M, Garg M. Phosphogypsumy ash cementitious binders its hydration
materials. J Mater Civil Eng 2001;5(13):34955. doi:10.1061/(ASCE)0899-
and strength development. Cem Concr Res 1995;4(25):7528. doi:10.1016/
1561(2001)13:5(349).
0008-8846(95)00065-K.
[25] olak A. The long-term durability performance of gypsumPortland cement
[10] Feng Q et al. Observation and analysis of gypsum particleboard using SEM. J
natural pozzolan blends. Cem Concr Res 2002;1(32):10915. doi:10.1016/
Wuhan Univ Technol Mater Sci Ed 2007;1(22):447. doi:10.1007/s11595-
S0008-8846(01)00637-8.
005-1044-.
[26] olak A. Physical and mechanical properties of polymerplaster composites.
[11] Singh M, Garg M. Investigation of a durable gypsum binder for building
Mater Lett 2006;16(60):197782. doi:10.1016/j.matlet.2005.12.062.
materials. Constr Build Mater 1992;1(6):526.
[27] Shen W et al. Study on limey ashphosphogypsum binder. Constr Build
[12] Bentur A, Klover K, Goldman A. Gypsum of improved performance using blends
Mater 2007;7(21):14805. doi:10.1016/j.conbuildmat.2006.07.010.
with Portland cement and silica fume. Adv Cem Res 1994;6(23):10916.
[28] De Milito, Performance evaluation of hemihydrateblastfurnace slag cement
[13] Singh M, Garg M. Relationship between mechanical properties and porosity of
silica fume binders. PhD Thesis, University of Campinas, Brazil; 2007. 340 p.
water-resistant gypsum binder. Cem Concr Res 1996;26(3):44956.
<http://www.unicamp.br/anuario/2007/FEC/FEC-tesesdoutorado.html> [in
doi:10.1016/S0008-8846(96)85032-0.
Portuguese].
[14] Alsadi S et al. Properties of gypsum with the addition of gum Arabic and
[29] Demir I, Baspinar MS. Effect of silica fume and expanded perlite addition on
calcium hydroxide. J Prosthet Dent 1996;76(5):5304. doi:10.1016/S0022-
technical properties of the y ashlimegypsum mixture. Constr Build Mater
3913(96)90013-X.
2008;6(22):1299304. doi:10.1016/j.conbuildmat.2007.01.011.
[15] Roy A et al. Microstructural and phase characteristics of phosphogypsum
[30] Santos RAM. Communicational dynamics registered by the members of the
cement mixtures. J Mater Civil Eng 1996;8(1):117. doi:10.1061/(ASCE)0899-
Black Alliance Possession and Cultural Center Active Force. PhD thesis.
1561(1996)8:1(11).
University of So Paulo, School of Communications and Arts; 2006. 294 p.
[16] Dharakumar M et al. Studies on Polymer Impregnated Plaster of Paris. J Appl
<http://www.teses.usp.br/teses/disponiveis/27/27134/tde-04102006-172027/
Polym Sci 1997;10(63):12439. doi:10.1002/(SICI)1097-4628(19970307)63:
pt-br.php> [in portuguese].
1061243::AID-APP1P3.0.CO;2-K.